%%HTML
<script src="require.js"></script>

%matplotlib inline
import math
import numpy as np, pandas as pd
import matplotlib.pyplot as plt, seaborn as sns
import plotly.express as px
import plotly.graph_objects as go
import plotly.io as pio
pio.renderers.default='notebook'

Importation des 3 fichiers CSV

df_games = pd.read_csv('data/games.csv', lineterminator='\n')
df_games.drop(columns=['summary'], inplace=True)

df_games['meta_count'] = (df_games['meta_pos'] + df_games['meta_mixed']+ df_games['meta_neg'])
df_games['user_count'] = (df_games['user_pos'] + df_games['user_mixed']+ df_games['user_neg'])
df_games['ratio'] = df_games['meta_count'] / df_games['user_count'] 
df_games['offset'] = df_games['meta_score'] - df_games['n_user_score']

On construit 4 colonnes de plus :

• meta_count pour toutes les critiques
• user_count pour tous les avis utilisateurs
• ratio pour le rapport entre les deux
• offset pour exprimer la différence entre la note meta et la note utilisateur

df_meta=pd.read_csv('data/meta_reviews.csv',lineterminator='\n')
df_users=pd.read_csv('data/user_reviews.csv',lineterminator='\n')

df_games.shape, df_meta.shape, df_users.shape

((31147, 22), (163577, 6), (176912, 6))

df_games.head()

	title	platform	developer	genre	rating	release_date	meta_score	meta_overview	meta_pos	meta_mixed	...	user_overview	user_pos	user_mixed	user_neg	n_user_score	real_date	meta_count	user_count	ratio	offset
0	007 Legends	Wii U	Eurocom Entertainment Software, Eurocom	['Action', 'Shooter', 'Shooter', 'First-Person...	T	Dec 11, 2012	0	No score yet	0	0	...	Mixed or average reviews	10.0	7.0	13.0	53.0	2012-12-11	3	30.0	0.10000	-53.0
1	1001 Spikes	Wii U	Nicalis	['Action', 'General', 'Platformer', 'Platforme...	T	Jun 3, 2014	83	Generally favorable reviews	5	1	...	Mixed or average reviews	14.0	3.0	6.0	70.0	2014-06-03	6	23.0	0.26087	13.0
2	140	Wii U	Jeppe Carlsen	['Action', 'Platformer', '2D']	E	Sep 1, 2016	0	No score yet	0	3	...	No user score yet	0.0	3.0	0.0	0.0	2016-09-01	3	3.0	1.00000	0.0
3	153 Hand Video Poker	Wii U	Skunk Software	['Miscellaneous', 'Gambling']	T	Jun 2, 2016	0	No score yet	0	0	...	No user score yet	0.0	0.0	1.0	0.0	2016-06-02	1	1.0	1.00000	0.0
4	360 Breakout	Wii U	nuGAME	['Action', 'Arcade']	E	Aug 11, 2016	0	No score yet	0	1	...	No user score yet	0.0	1.0	0.0	0.0	2016-08-11	1	1.0	1.00000	0.0

5 rows × 22 columns

On crée df_games2 qui ne contient que les jeux ayant à la fois une note globale utilisateur et une note globale de critique

df_games2 = df_games.loc[(df_games['meta_overview'] != 'No score yet') & (df_games['user_overview'] != 'No user score yet') & (df_games['user_count'] != 0)]
df_games2.head()

	title	platform	developer	genre	rating	release_date	meta_score	meta_overview	meta_pos	meta_mixed	...	user_overview	user_pos	user_mixed	user_neg	n_user_score	real_date	meta_count	user_count	ratio	offset
1	1001 Spikes	Wii U	Nicalis	['Action', 'General', 'Platformer', 'Platforme...	T	Jun 3, 2014	83	Generally favorable reviews	5	1	...	Mixed or average reviews	14.0	3.0	6.0	70.0	2014-06-03	6	23.0	0.260870	13.0
5	3Souls	Wii U	Red Column	['Action', 'Action Adventure', 'Platformer', '...	E10+	Jul 28, 2016	49	Generally unfavorable reviews	0	3	...	Mixed or average reviews	4.0	2.0	1.0	69.0	2016-07-28	4	7.0	0.571429	-20.0
12	A World of Keflings	Wii U	NinjaBee	['Strategy', 'Breeding/Constructing', 'General...	E10+	Nov 13, 2014	70	Mixed or average reviews	3	1	...	Generally favorable reviews	5.0	1.0	1.0	76.0	2014-11-13	5	7.0	0.714286	-6.0
14	Abyss	Wii U	EnjoyUp Games	['Action', 'General', 'Platformer', '2D']	E	May 1, 2014	59	Mixed or average reviews	0	6	...	Mixed or average reviews	5.0	5.0	2.0	65.0	2014-05-01	7	12.0	0.583333	-6.0
18	Adventure Time: Explore the Dungeon Because I ...	Wii U	WayForward	['Action Adventure', 'Fantasy', 'General', 'Fa...	E10+	Nov 19, 2013	40	Generally unfavorable reviews	0	2	...	Mixed or average reviews	10.0	5.0	11.0	57.0	2013-11-19	7	26.0	0.269231	-17.0

5 rows × 22 columns

px.histogram(df_games2,x='meta_score')

px.histogram(df_games2,x='offset')

On affiche la distribution des ratios "nombre de critiques / nombre de revues utilisatuer". On voit que la grande majorité des ratio est inférieure à 1, montrant qu'il y a plus de participation des utilisateurs. On pourrait parler du pic de la distribution (du mode) pour avoir une idée du ratio moyen.

px.histogram(df_games2,x='ratio',template="simple_white")

Ici un graphe qui montre la distribution des scores meta pour chaque console. Ça se ressemble. Petite tendance qui se dégage : les consoles les plus récentes (Series X et PS5) sont plus distribuées vers les notes hautes. (Pour information, la limite droite de chaque boite indique le 3ème quartile, c'est à dire que 25% des notes sont supérieures)

fig = px.box(df_games2,x='meta_score',color='platform',hover_name='title',
             title="Distribution des notes Meta par plateforme",
             category_orders={
                "platform": ["3DS", "Wii U", "Switch", "Playstation 4", "Playstation 5","Xbox One","Xbox Series X"],
            },
            template="simple_white")
fig.show()

df =pd.DataFrame(dict(
    series=np.concatenate((["Meta Score"]*len(df_games2['meta_score']), ["User Score"]*len(df_games2['n_user_score']))), 
    data  =np.concatenate((df_games2['meta_score'],df_games2['n_user_score']))
))

px.histogram(df, x="data", color="series", barmode="overlay",template="simple_white")

fig=px.scatter(df_games2,x='meta_score',y='n_user_score',
               color='platform', hover_data=['title'],size='meta_count',trendline='ols',template="simple_white")
fig.add_shape(type="line",
    x0=0, y0=0, x1=100, y1=100,
    line=dict(color="RoyalBlue",width=3)
)
fig.update_layout(width=1000, height=800)

fig.show()

df_meta.head().transpose()

	0	1	2	3	4
score	42.0	40.0	40.0	95.0	90.0
critic	IGN	Official Nintendo Magazine UK	Nintendo Life	Destructoid	FNintendo
date	Jan 14, 2013	Jan 27, 2013	Dec 17, 2012	Jun 8, 2014	Nov 5, 2015
title	007 Legends	007 Legends	007 Legends	1001 Spikes	1001 Spikes
platform	Wii U	Wii U	Wii U	Wii U	Wii U
real_date	2013-01-14	2013-01-27	2012-12-17	2014-06-08	2015-11-05

# si j'utilisais pyplot
plt.rc('xtick', labelsize=14)
plt.rc('ytick', labelsize=14)
sns.set_style(style='whitegrid')

General

Which console has more games titles ?

df_games.groupby('platform')['title'].count().sort_values(ascending=False).plot.bar(figsize=(10, 6), rot=0);

Je regarde combien de critiques sont faites par chaque organe identifié dans Metacritique. On comprend ici que TheXboxHub a produit 3923 critiques, Nintendo Life 3864...

fig = px.histogram(df_games2, x = 'meta_score')
fig.show()

df_meta.groupby('critic')['score'].count()
#.sort_values(ascending=False)[:1000]
#.plot.bar(figsize=(10, 6), rot=0);

critic
1UP                   20
3DJuegos            1151
4Players.de         1770
ActionTrip            99
Adventure Gamers      26
                    ... 
XboxEra              164
Yahoo!                59
ZTGD                1702
games(TM)            654
iMore                 54
Name: score, Length: 327, dtype: int64

df = df_meta[df_meta['critic'] == 'TheXboxHub']
#df_meta.groupby('critic')['score'].count().sort_values(ascending=False)[:25]
#.plot.bar(figsize=(10, 6), rot=0);
fig = px.histogram(df_meta, x="score",marginal='box')
fig.show()

Given release date let's add two columns with year and month information

df_meta[df_meta['critic'] == 'TheXboxHub']

	score	critic	date	title	platform	real_date
66609	70.0	TheXboxHub	Mar 10, 2021	3 out of 10: Season One	Xbox Series X	2021-03-10
66612	30.0	TheXboxHub	Oct 31, 2022	41 Hours	Xbox Series X	2022-10-31
66613	80.0	TheXboxHub	Dec 3, 2021	6Souls	Xbox Series X	2021-12-03
66614	40.0	TheXboxHub	Nov 16, 2022	7 Days of Rose	Xbox Series X	2022-11-16
66616	60.0	TheXboxHub	Oct 3, 2022	9 Clues 2: The Ward	Xbox Series X	2022-10-03
...	...	...	...	...	...	...
163513	60.0	TheXboxHub	Nov 14, 2019	Zombieland: Double Tap - Road Trip	Xbox One	2019-11-14
163516	80.0	TheXboxHub	Jul 14, 2021	Zombies Ate My Neighbors and Ghoul Patrol	Xbox One	2021-07-14
163518	60.0	TheXboxHub	May 10, 2020	Zombies ruined my day	Xbox One	2020-05-10
163519	70.0	TheXboxHub	Jun 20, 2022	Zombo Buster Advance	Xbox One	2022-06-20
163520	50.0	TheXboxHub	Dec 1, 2021	Zombo Buster Rising	Xbox One	2021-12-01

3923 rows × 6 columns

df['release_date'] = pd.to_datetime(df['release_date'])
df['month'] = df['release_date'].dt.month
df['year'] = df['release_date'].dt.year

EA Sports appears also as 'EA Sports, EA Vancouver' and 'EA Vancouver'. Let's unify the names

df.loc[df['developer'] == 'EA Sports, EA Vancouver', 'developer'] = 'EA Sports'
df.loc[df['developer'] == 'EA Vancouver', 'developer'] = 'EA Sports'

In order to perform analysis related to scores, we need to use those games with reviews

tmp = df.loc[(df['meta_overview'] != 'No score yet') & (df['user_overview'] != 'No user score yet')].copy()

Which console has best average scores ?

tmp.groupby('platform')['meta_score', 'n_user_score'].mean().sort_values('meta_score', ascending=False).plot.bar(figsize=(10, 6), rot=0);

/var/folders/2h/pylp0hzn5td4mz63q7r0cz0h0000gn/T/ipykernel_30894/3120562876.py:1: FutureWarning: Indexing with multiple keys (implicitly converted to a tuple of keys) will be deprecated, use a list instead.
  tmp.groupby('platform')['meta_score', 'n_user_score'].mean().sort_values('meta_score', ascending=False).plot.bar(figsize=(10, 6), rot=0);

Let's compare metacritics scores vs user scores. (metacritics scores are made by specialized critics sites while user scores by people registered on metacritic.com)

tmp[['meta_score']].plot.hist(bins=90, figsize=(12, 6));

tmp[['n_user_score']].plot.hist(bins=25, figsize=(12, 6));

Meta scores tend to be higher than user scores

fig, ax = plt.subplots(figsize=(12, 8))
ax.plot([0, 100], [0, 100])
tmp.plot.scatter(x='n_user_score', y='meta_score', ax=ax, alpha=0.25);

mo = tmp.groupby('meta_overview')['title'].count()
mo.name = 'meta_overview'
uo = tmp.groupby('user_overview')['title'].count()
uo.name = 'user_overview'
pd.concat([mo, uo], axis=1).sort_values('meta_overview', ascending=False).plot.bar(figsize=(12, 6));

We confirm that there are more 'Generally favorable reviews' in meta scores and more 'Generally unfavorable reviews' in user scores.
Users tend to be more critic in their reviews than specialized sites (paid reviews ?)

How many new titles were released since 2014 ?

Here we compare just PlayStation 4 and Xbox One because both platforms were announced the same year (2013). On the other hand Nintendo Switch in 2016

# df.loc[(df['year'] > 2013) & (df['year'] < 2023) & (df['platform'] != 'Switch')].groupby(['year', 'platform'])['title'].count().unstack().plot.bar(figsize=(10, 6));
df.loc[(df['year'] > 2011) & (df['year'] < 2023)].groupby(['year', 'platform'])['title'].count().unstack().plot.line(figsize=(15, 6));

When are new titles released ?

df.groupby(['year', 'month'])['title'].count().groupby('month').mean().plot(figsize=(12, 6));

There's a peak between August and November (anticipating christmas sales?)

How did average scores evolve since 2014 ?

fig = plt.figure(figsize = (15,6))
ax = fig.add_subplot(1, 1, 1)
tmp.loc[(tmp['year'] > 2012) & (tmp['year'] < 2023), ('meta_score', 'year')].groupby('year').mean().plot(ax = ax, xticks=np.arange(2012, 2022, 1))
tmp.loc[(tmp['year'] > 2012) & (tmp['year'] < 2023), ('n_user_score', 'year')].groupby('year').mean().plot(ax = ax, xticks=np.arange(2012, 2022, 1));

No big changes

Best games

How many titles per console have metacritic score greater than 85?

tmp.loc[tmp['meta_score'] >= 85].groupby('platform')['title'].count().sort_values(ascending=False)

platform
PlayStation 4    188
Switch           172
Xbox One         158
PlayStation 5     52
Xbox Series X     52
3DS               40
Wii U             31
Name: title, dtype: int64

How many titles per console have user score greater than 85?

tmp.loc[tmp['n_user_score'] >= 85].groupby('platform')['title'].count().sort_values(ascending=False)

platform
PlayStation 4    86
Switch           73
Xbox One         47
Xbox Series X    33
Wii U            23
3DS              20
PlayStation 5    16
Name: title, dtype: int64

Let's add a column with the difference between meta and user score

tmp['dif'] = tmp['meta_score'] - tmp['n_user_score']

ps4 = tmp.loc[tmp['platform'] == 'PlayStation 4']
xone = tmp.loc[tmp['platform'] == 'Xbox One']
switch = tmp.loc[tmp['platform'] == 'Switch']

PlayStation 4

What are the best 20 titles according to metacritic score?

ps4.sort_values('meta_score', ascending=False).reset_index(drop=True)[['title', 'meta_score', 'n_user_score', 'dif']][:20]

	title	meta_score	n_user_score	dif
0	Grand Theft Auto V	97	83.0	14.0
1	Red Dead Redemption 2	97	87.0	10.0
2	Persona 5 Royal	95	85.0	10.0
3	The Last of Us Remastered	95	92.0	3.0
4	God of War	94	91.0	3.0
5	The Last of Us Part II	93	58.0	35.0
6	Metal Gear Solid V: The Phantom Pain	93	82.0	11.0
7	Persona 5	93	87.0	6.0
8	XCOM 2: War of the Chosen	93	69.0	24.0
9	Uncharted 4: A Thief's End	93	88.0	5.0
10	Journey	92	83.0	9.0
11	Divinity: Original Sin II - Definitive Edition	92	80.0	12.0
12	Undertale	92	73.0	19.0
13	Bloodborne	92	89.0	3.0
14	The Witcher 3: Wild Hunt	92	92.0	0.0
15	The Witcher 3: Wild Hunt - Blood and Wine	91	92.0	-1.0
16	Dragon Quest XI S: Echoes of an Elusive Age - ...	91	80.0	11.0
17	Resident Evil 2	91	90.0	1.0
18	Shadow of the Colossus	91	78.0	13.0
19	Final Fantasy XIV: Shadowbringers	91	87.0	4.0

For a lot of those titles there's a big difference between what meta critics and users think

What are the best 20 titles according to user score?

ps4.sort_values('n_user_score', ascending=False).reset_index(drop=True)[['title', 'n_user_score', 'meta_score', 'dif']][:20]

	title	n_user_score	meta_score	dif
0	Wanderer	98.0	72	-26.0
1	Dynasty Warriors 9 Empires	96.0	58	-38.0
2	Crystar	96.0	67	-29.0
3	Bayonetta & Vanquish	93.0	85	-8.0
4	Star Renegades	93.0	78	-15.0
5	Fatal Frame: Maiden of Black Water	93.0	71	-22.0
6	The Witcher 3: Wild Hunt	92.0	92	0.0
7	The Witcher 3: Wild Hunt - Blood and Wine	92.0	91	-1.0
8	Infinifactory	92.0	70	-22.0
9	The Evil Within: The Assignment	92.0	80	-12.0
10	The Last of Us Remastered	92.0	95	3.0
11	God of War	91.0	94	3.0
12	Ghost of Tsushima	91.0	83	-8.0
13	The Evil Within: The Consequence	91.0	75	-16.0
14	Vanquish	91.0	77	-14.0
15	Dark Souls III	90.0	89	-1.0
16	Astro Bot: Rescue Mission	90.0	90	0.0
17	Atelier Sophie 2: The Alchemist of the Mysteri...	90.0	82	-8.0
18	Rise of the Third Power	90.0	79	-11.0
19	The Witcher 3: Wild Hunt - Hearts of Stone	90.0	90	0.0

With exception of AO Tennis, meta critics and users scores seem to be similar

What are the top 10 titles most overrated by metacritic? (user point of view)

ps4.sort_values('dif', ascending=False)[:20][['title', 'meta_score', 'n_user_score', 'dif']].reset_index(drop=True)

	title	meta_score	n_user_score	dif
0	ZEN Pinball 2: Ant-Man Pinball	81	12.0	69.0
1	FIFA 20	79	13.0	66.0
2	NBA 2K20	78	14.0	64.0
3	FIFA 21	72	9.0	63.0
4	Tom Clancy's The Division 2: Warlords of New York	74	11.0	63.0
5	Madden NFL 20	76	14.0	62.0
6	NBA 2K18	80	18.0	62.0
7	Tales of Symphonia Remastered	70	8.0	62.0
8	Megaquarium	78	16.0	62.0
9	FIFA 19	83	22.0	61.0
10	Madden NFL 21	63	3.0	60.0
11	EA SPORTS UFC 4	78	18.0	60.0
12	Madden NFL 19	80	20.0	60.0
13	Call of Duty: Infinite Warfare - Sabotage	76	17.0	59.0
14	NBA 2K21	68	9.0	59.0
15	Gran Turismo 7	82	26.0	56.0
16	Call of Duty: WWII - The War Machine	73	17.0	56.0
17	NBA 2K19	82	28.0	54.0
18	Call of Duty: Black Ops III - Salvation	74	21.0	53.0
19	Call of Duty: Infinite Warfare - Continuum	75	22.0	53.0

What are the top 10 titles most underrated by metacritic? (user point of view)

ps4.sort_values('dif')[:10][['title', 'meta_score', 'n_user_score', 'dif']].reset_index(drop=True)

	title	meta_score	n_user_score	dif
0	Left Alive	37	82.0	-45.0
1	Dynasty Warriors 9 Empires	58	96.0	-38.0
2	Dead or Alive Xtreme 3: Fortune	43	75.0	-32.0
3	Crystal Rift	37	66.0	-29.0
4	Crystar	67	96.0	-29.0
5	AO Tennis	44	73.0	-29.0
6	Big Bash Boom	55	83.0	-28.0
7	Terminator: Resistance	47	75.0	-28.0
8	Energy Hook	22	49.0	-27.0
9	Spuds Unearthed	43	70.0	-27.0

Let's see more information about AO Tennis

ps4.loc[ps4['title'] == 'AO Tennis'].transpose()

	1242
title	AO Tennis
platform	PlayStation 4
developer	Big Ant Studios
genre	['Sports', 'Individual', 'Tennis']
rating	NaN
release_date	2018-01-16 00:00:00
meta_score	44
meta_overview	Generally unfavorable reviews
meta_pos	0
meta_mixed	2
meta_neg	2
user_score	7.3
user_overview	Mixed or average reviews
user_pos	28.0
user_mixed	1.0
user_neg	11.0
n_user_score	73.0
real_date	2018-01-16
month	1
year	2018
dif	-29.0

It's a new game with few critics reviews

Exclusive games

ps4_list = set(df.loc[df['platform'] == 'PlayStation 4', 'title'])
xone_list = set(df.loc[df['platform'] == 'Xbox One', 'title'])
switch_list = set(df.loc[df['platform'] == 'Switch', 'title'])

PlayStation 4

How many exclusive games PlayStation 4 got?

ps4_exclusives = tmp.loc[tmp['title'].isin(list(ps4_list.difference(xone_list).difference(switch_list))), ('title', 'n_user_score', 'meta_score')].sort_values('n_user_score', ascending=False).reset_index(drop=True)
len(ps4_exclusives)

700

What are the top 25 exclusives PlayStation 4 games?

ps4_exclusives[:25]

	title	n_user_score	meta_score
0	Wanderer	98.0	72
1	Infinifactory	92.0	70
2	The Last of Us Remastered	92.0	95
3	Ghost of Tsushima	91.0	83
4	God of War	91.0	94
5	Astro Bot: Rescue Mission	90.0	90
6	NieR: Automata	89.0	88
7	Bloodborne: The Old Hunters	89.0	87
8	Bloodborne	89.0	92
9	NieR: Automata - Game of the YoRHa Edition	89.0	91
10	Detroit: Become Human	88.0	78
11	Uncharted 4: A Thief's End	88.0	93
12	Utawarerumono: Mask of Truth	88.0	75
13	Final Fantasy XIV: Stormblood	88.0	89
14	Final Fantasy XIV: Endwalker	87.0	90
15	Marvel's Spider-Man	87.0	87
16	Persona 5	87.0	93
17	Dreams	87.0	89
18	Final Fantasy XIV: Shadowbringers	87.0	91
19	Monster of the Deep: Final Fantasy XV	87.0	62
20	Uncharted: The Nathan Drake Collection	86.0	86
21	The Legend of Heroes: Trails of Cold Steel	86.0	80
22	Sackboy: A Big Adventure	86.0	79
23	Dragon Quest XI: Echoes of an Elusive Age	86.0	86
24	Heavy Rain & Beyond: Two Souls Collection	85.0	78

How many exlusive PlayStation 4 games have user score equal or greater than 80 ?

len(ps4_exclusives.loc[ps4_exclusives['n_user_score'] >= 80])

87

Xbox One

How many exclusive games Xbox One got?

xone_exclusives = tmp.loc[tmp['title'].isin(list(xone_list.difference(ps4_list).difference(switch))), ('title', 'n_user_score', 'meta_score')].sort_values('n_user_score', ascending=False).reset_index(drop=True)
len(xone_exclusives)

325

What are the top 25 exclusives Xbox One games?

xone_exclusives[:25]

	title	n_user_score	meta_score
0	NieR: Automata - Become as Gods Edition	91.0	90
1	Ori and the Will of the Wisps	89.0	90
2	Syberia: The World Before	88.0	82
3	Ori and the Will of the Wisps	87.0	92
4	Ori and the Will of the Wisps	87.0	93
5	Ori and the Blind Forest	86.0	88
6	Age of Empires II: Definitive Edition	86.0	86
7	Katana ZERO	85.0	83
8	Ori and the Blind Forest: Definitive Edition	85.0	90
9	Floppy Knights	85.0	69
10	Vampire Survivors	85.0	95
11	Octopath Traveler	84.0	83
12	Enter The Gungeon	83.0	85
13	Omensight: Definitive Edition	83.0	81
14	Syberia: The World Before	83.0	77
15	Tunic	83.0	88
16	Dying Light: Platinum Edition	83.0	78
17	Forza Horizon 4	83.0	92
18	Forza Horizon 2	83.0	86
19	Tunche	82.0	71
20	MouseCraft	82.0	77
21	The Binding of Isaac: Afterbirth +	82.0	85
22	Turok 2: Seeds of Evil Remaster	82.0	78
23	The Last Oricru	82.0	57
24	Forza Horizon 5	82.0	92

How many exlusive Xbox One games have user score equal or greater than 80 ?

len(xone_exclusives.loc[xone_exclusives['n_user_score'] >= 80])

44

Switch

How many exclusive games Switch got?

switch_exclusives = tmp.loc[tmp['title'].isin(list(switch_list.difference(ps4_list).difference(xone_list))), ('title', 'n_user_score', 'meta_score')].sort_values('n_user_score', ascending=False).reset_index(drop=True)
len(switch_exclusives)

666

What are the top 25 exclusives Switch games?

switch_exclusives[:25]

	title	n_user_score	meta_score
0	Ultra Kaiju Monster Rancher	95.0	75
1	Touken Ranbu Warriors	92.0	68
2	Monster Rancher 1 & 2 DX	90.0	70
3	Bayonetta 2	89.0	91
4	Fire Emblem Warriors: Three Hopes	89.0	81
5	Donkey Kong Country: Tropical Freeze	89.0	83
6	Astral Chain	89.0	87
7	Super Mario Odyssey	89.0	97
8	Kirby and the Forgotten Land	88.0	85
9	Xenoblade Chronicles: Definitive Edition	88.0	89
10	Fire Emblem: Three Houses	88.0	89
11	Metroid Prime Remastered	88.0	94
12	Picross S 4	87.0	77
13	Bayonetta + Bayonetta 2	87.0	90
14	Super Mario 3D World + Bowser's Fury	87.0	89
15	The Legend of Zelda: Breath of the Wild	87.0	97
16	Captain Toad: Treasure Tracker	86.0	81
17	Ring Fit Adventure	86.0	83
18	The House in Fata Morgana - Dreams of the Reve...	86.0	96
19	Donkey Kong Country: Tropical Freeze	86.0	86
20	Pikmin 3 Deluxe	86.0	85
21	Mario Kart 8 Deluxe	86.0	92
22	Bayonetta 2	86.0	92
23	Metroid Dread	86.0	88
24	Monster Train First Class	85.0	85

How many exlusive Switch games have user score equal or greater than 80 ?

len(switch_exclusives.loc[switch_exclusives['n_user_score'] >= 80])

146

Developers

PlayStation 4

Who are the top 10 developers with more released titles?

ps4.groupby('developer')['title'].count().sort_values(ascending=False)[:10]

developer
Telltale Games          53
Capcom                  41
Square Enix             31
Omega Force             23
Ubisoft Montreal        21
EA Sports               20
Zen Studios             20
Bandai Namco Games      18
Milestone S.r.l         16
Nippon Ichi Software    16
Name: title, dtype: int64

Does quantity mean quality?

dev = pd.concat([ps4.groupby('developer')['title'].count(), ps4.groupby('developer')['n_user_score'].mean(), ps4.groupby('developer')['meta_score'].mean()], axis=1)
dev.columns = ['count', 'n_user_score_avg', 'meta_score_avg']
dev.sort_values('count', ascending=False)[:10]

	count	n_user_score_avg	meta_score_avg
developer
Telltale Games	53	65.811321	73.150943
Capcom	41	70.121951	76.195122
Square Enix	31	76.580645	76.290323
Omega Force	23	74.000000	71.826087
Ubisoft Montreal	21	66.428571	71.761905
EA Sports	20	45.800000	73.550000
Zen Studios	20	54.200000	78.350000
Bandai Namco Games	18	58.611111	68.666667
Milestone S.r.l	16	60.312500	69.625000
Nippon Ichi Software	16	66.000000	70.625000

It seems that not

Who are the best 10 developers according to user score (at least 4 released titles)?

dev.loc[dev['count'] >= 4].sort_values('n_user_score_avg', ascending=False)[:10]

	count	n_user_score_avg	meta_score_avg
developer
CD Projekt Red Studio	4	86.750000	88.250000
Tango Gameworks	5	83.000000	73.000000
From Software	9	83.000000	83.333333
Falcom	10	81.200000	79.300000
Arkane Studios	5	81.000000	82.400000
Ryu ga Gotoku Studios	10	79.800000	80.400000
Naughty Dog	4	79.500000	91.250000
id Software	6	77.833333	78.666667
Grasshopper Manufacture	4	77.250000	66.750000
PlatinumGames	10	76.600000	75.200000

What are the titles that gave them such a good reputation ?

From Software

ps4.loc[ps4['developer'] == 'From Software', 'title'].reset_index(drop=True)

0                                 Bloodborne
1                Bloodborne: The Old Hunters
2    Dark Souls II: Scholar of the First Sin
3                             Dark Souls III
4          Dark Souls III: Ashes of Ariandel
5            Dark Souls III: The Ringed City
6                      Dark Souls Remastered
7                                   Deracine
8                  Sekiro: Shadows Die Twice
Name: title, dtype: object

DONTNOD Entertainment

ps4.loc[ps4['developer'] == 'DONTNOD Entertainment', 'title'].reset_index(drop=True)

0                               Life is Strange
1                             Life is Strange 2
2          Life is Strange 2: Episode 1 - Roads
3          Life is Strange 2: Episode 2 - Rules
4     Life is Strange 2: Episode 3 - Wastelands
5          Life is Strange 2: Episode 4 - Faith
6         Life is Strange 2: Episode 5 - Wolves
7        Life is Strange: Episode 1 - Chrysalis
8      Life is Strange: Episode 2 - Out of Time
9     Life is Strange: Episode 3 - Chaos Theory
10       Life is Strange: Episode 4 - Dark Room
11       Life is Strange: Episode 5 - Polarized
12     The Awesome Adventures of Captain Spirit
13                                       Vampyr
Name: title, dtype: object

Who are the top 10 developers whose titles are on average more overrated by metacritic (at least 4 released titles) ?

dev = pd.concat([ps4.groupby('developer')['dif'].mean().sort_values(ascending=False), ps4.groupby('developer')['title'].count()], axis=1)
dev.columns = ['dif_avg', 'count']
dev.loc[dev['count'] >= 4].sort_values('dif_avg', ascending=False)[:10]

	dif_avg	count
developer
Electronic Arts	55.142857	7
Sledgehammer Games	39.800000	5
Visual Concepts	38.500000	10
Treyarch	33.714286	7
Infinity Ward	33.555556	9
Bungie	30.333333	9
EA Sports	27.750000	20
Massive Entertainment	25.250000	4
Zen Studios	24.150000	20
EA DICE	23.375000	8

What are the top 10 EA Sports games according to metacritic?

ea = ps4.loc[ps4['developer'] == 'EA Sports', ['title', 'meta_score', 'n_user_score']].sort_values('meta_score', ascending=False)[:10]
ea.set_index('title').plot.bar(figsize=(14, 6));

Who are the top 10 developers whose titles are on average more underrated by metacritic (at least 4 released titles)?

dev.loc[dev['count'] >= 4].sort_values('dif_avg')[:10]

	dif_avg	count
developer
Grasshopper Manufacture	-10.500000	4
Tango Gameworks	-10.000000	5
Big Ant Studios	-8.500000	8
Dimps Corporation	-7.800000	5
CyberConnect2	-4.571429	7
Koei Tecmo Games	-4.538462	13
Acquire	-3.833333	6
Omega Force	-2.173913	23
Cyanide, Cyanide Studios	-2.000000	4
Falcom	-1.900000	10

ps4.loc[ps4['developer'] == 'Grasshopper Manufacture', 'title'].reset_index(drop=True)

0                                           Let It Die
1                       The 25th Ward: The Silver Case
2                                      The Silver Case
3    Travis Strikes Again: No More Heroes - Complet...
Name: title, dtype: object

ps4.loc[ps4['developer'] == 'Acquire', 'title'].reset_index(drop=True)

0         Aegis of Earth: Protonovus Assault
1                               Akiba's Beat
2        Akiba's Trip: Hellbound & Debriefed
3           Akiba's Trip: Undead & Undressed
4    Katana Kami: A Way of the Samurai Story
5                      No Heroes Allowed! VR
Name: title, dtype: object

Ratings

How many titles per rating category?

df['rating'].value_counts().plot.bar(figsize=(12, 6), rot=0);

E: Everyone
E+10: Everyone +10
T: Teen +13
M: Mature +17

Which console is more "family" oriented?

qty = df.loc[(df['rating'].isnull() == False) & (df['rating'] != 'RP')].groupby('platform')['title'].count()
#((df.loc[df['rating'] != 'RP'].groupby(['platform', 'rating'])['title'].count().unstack().transpose() / qty) * 100).plot.bar(figsize=(12, 6), rot=0);
rat = (df.loc[df['rating'] != 'RP'].groupby(['platform', 'rating'])['title'].count().unstack().transpose() / qty) * 100
rat = rat[['PlayStation 4', 'Xbox One', 'Switch']]

fig = plt.figure(figsize = (10,6))
ax = fig.add_subplot(1, 1, 1)
rat.iloc[0].plot.bar(ax=ax, color='beige', edgecolor='white', label=rat.index[0]);
rat.iloc[1].plot.bar(ax=ax, bottom=rat.iloc[0], color='navajowhite', edgecolor='white', label=rat.index[1]);
rat.iloc[3].plot.bar(ax=ax, bottom=rat.iloc[:2].sum().values, color='lightsalmon', edgecolor='white', label=rat.index[3]);
rat.iloc[2].plot.bar(ax=ax, bottom=(rat.iloc[:2].sum() + rat.iloc[3]).values, color='firebrick', edgecolor='white', label=rat.index[2], rot=0);
ax.legend(loc='upper left', bbox_to_anchor=(1, 1), ncol=1);

Switch has in proportion more games suited for everyone

Which category has the best games ?

tmp.loc[(tmp['rating'].isnull() == False) & (tmp['rating'] != 'RP')].groupby('rating')['n_user_score'].mean().sort_values(ascending=False).plot.bar(figsize=(12, 6), rot=0);

Mature category seems to has an advantage